Prof. Naomi Rothenberg (University of Memphis)

ABSTRACT: This paper studies how performance measurement bias affects the principal’s choice of whether pre-decision information about the productivity environment should be restricted to one agent or whether both agents should have access. Agents work sequentially and in close environments so that agents can observe each other’s effort. If only the upstream agent is privately informed, information can be signaled to the downstream agent through the choice of productive action, but if both agents are informed, the downstream agent has no incentive to learn from observing the upstream agent’s effort. With individual measures of performance, the choice of whether one or both agents should be privately informed is affected by the performance measurement error. With both agents informed, using relative performance evaluation is optimal, which reduces agents’ information rents. With only the upstream agent informed, joint performance evaluation can also be optimal due to the effect of the uninformed agent’s actions on the upstream agent’s incentives. The performance measurement error affects the informativeness of the performance measure, depending on whether good or bad output is reported and depending on whether relative or joint performance evaluation is optimal. Increasing the precision of the performance measure or the significance of the decision-making function of the private information makes it more likely the principal prefers both agents equally informed.